Meta’s Muse Spark Ranks No. 2 on OpenCode’s Model Statistics Page
The model recorded 36 trillion displayed tokens and an 83.3% weekly-retention rate among 13,000 eligible users on Sept. 24.
Meta’s Muse Spark 1.3 Contributor ranked No. 2 in recent usage on OpenCode’s public model-statistics page Sept. 24, recording 36 trillion displayed tokens and signaling strong developer demand for the discounted model.
The model also ranked No. 2 in weekly retention, with an 83.3% rate among 13,000 eligible users. Meta models collectively accounted for 4.1 trillion tokens and 23.7% of the displayed market share.
OpenCode Go is a subscription service that gives coding agents access to multiple artificial-intelligence models. Its model lineup includes offerings from Meta, DeepSeek, Xiaomi, NVIDIA and OpenAI, and is designed for OpenCode and other coding agents that generate similar requests.
Muse Spark 1.3 Contributor is priced at $0.10 per 1 million input tokens, $0.20 per 1 million output tokens and $0.002 per 1 million cached-read tokens. The monthly usage limit is $60.
The Contributor model is offered at heavily discounted prices in exchange for permission to use prompts and completions to train future Meta models. Availability is limited by Meta’s geographic-use policy.
The Sept. 24 statistics show token totals and retention data, but do not establish platform-wide session counts or total developer-token activity. The page also does not fully define its ranking methodology or measurement period, and its date listing does not specify a time zone.
Meta introduced Muse Spark on April 8 as the first model in a new series developed by Meta Superintelligence Labs. The model powers the Meta AI app and meta.ai, supports complex reasoning and multimodal tasks, and can create websites and mini-games through visual coding.
The initial model was designed to be small and fast. Its developer-service ranking adds a usage indicator for a model otherwise positioned across Meta’s consumer artificial-intelligence products.
